There are four key concepts that can help with effective searching: breaking up a search into its conceptual parts, using boolean logic to tie concepts together, using stemming and wildcards to account for variations in how terms are written or articulated, and ensuring that concepts that are comprised of more than one term are grouped together.
Research questions are generally comprised of at least 3 concepts: the organism or population of interest, the intervention or thing being studied in relation to that organism or population, and the measure of interest or the outcome that we’re interested in. This is often formalized as a PIO framework - Population, Intervention, Outcome. There are many variations on this, including a PEO – Population, Exposure, Outcome – PICO where the C covers a comparison – PICOS, where the S is for Study type etc. These frameworks can be useful for articulating a research question in terms of it’s conceptual components; pick the simplest one that resonates with you and your question.
Putting this into an example, we might ask about the relationship between smoking and cancer in men. We have a population – men – an intervention, or perhaps more appropriate an exposure – smoking – and an outcome – cancer. A very simple search for this question might then be:
men AND smoking AND cancer
Similarlily, we might ask if an increase in water salinity increases mortality in zebra fish. Again, we have a population – zebra fish – an intervetion or exposure (if an experimental study it would be an intervention, if an observational study it would be an exposure) – salinity – and an outcome to measure – mortality. A very simple search for this question might then be:
“zebra fish” AND salinity AND mortality
Boolean logical operators are comprised of AND OR and NOT. Generally when searching the literature, we don’t use NOT, but there are circumstances where it may be warranted.
AND is an intersect, OR is a union. AND means both elements must be in the set, OR means either element could be in the set. AND is thus used to find the intersection between concepts while OR is used to find the union among synonyms. In the example
men AND smoking AND cancer
AND is used to find results that have all three terms. Often times, there is more than one way to articualte a given concept, which is when we use OR. For example
men AND smoking AND (cancer OR neoplasm)
We always want to use a reasonable number of synonyms to ensure we are capturing different approaches to describing any given entity or phenomenon.
Words can have alternate spellings. There are two ways to work around this. We could use OR, for example
color OR colour
Alternatively, we could use a wildcard
where the # represents 1 or 0 characters.
Wildcards will differe between databases. Read the documentation to know which wildcards are available and how to use them.
Stemming stems a word. For example, we might want to find the term smoke, smoker, smoking, smoked etc. Again, we could use OR
smoke OR smoker OR smoking OR smoked
Alternatively, we could use a stem
which will look for the letters smok and any combination of letters thereafter, capturing all the variations (and more potentially) of interest.
The stemming wildcard * is generally pretty universal across databases. However, neither stemming nor wildcards more generally operate in Google or Google Scholar in the same way as they do in the more structured databases listed in the Sources of Evidence section.
When a concept is comprised of more than one term, we need to explicity account for that, and we do that with the use of quotations. For example, a search for
is equivalent to a search for
zebra AND fish
which is significantly broader than a search for
where, in the former, both terms need to be present, and in the latter they need to be not only present, but also directly adjacent to each other.